JPH0933216A - Image processor carried on vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an image processor by which a white line can be recognized stably without being subjected to the influence of a noise which is stable in a time-series manner by deciding the white line on the basis of traffic-lane-width information from an extracted white-line candidate point. SOLUTION: In a preprocessing means 28, an image in which a contour has been extracted and which has been binalized, is subjected to time quadrature by using integration device, and an output signal by a time-quardrature means 29 is input to a region division means 30. This is performed by using a comparison means. When the signal is larger than a specific threshold, a region I is divided into two regions on a screen, and, when the signal is small, a region O is divided into two regions. Then, a deletion processing operation such as a white-line recognition operation or a vehicle recognition operation is limited to the region I. Thereby, a processing speed can be increased as a whole. Then, by using the region I, a white line is recognized by a white-line recognition means 3. When the white line is recognized, the white line inside the region I is found from its contour. In this manner, a designated region is selected as a white-line candidate point, and the white line is decided according to every region.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle-mounted image processing apparatus, and more particularly to a vehicle-mounted image processing apparatus for performing white line recognition.

[0002]

2. Description of the Related Art Conventionally, as an automotive white line recognition device known as an in-vehicle image processing device, as shown in FIG. 10, an image signal is input from an image information input from a CCD camera 1 by an A / D conversion means 2. It is known that a white signal is converted into a digital signal and white line recognition is performed by the white line recognition means 3 from the digitally converted signal.

In the prior art, various white line recognizing means for obtaining a white line position by recognizing a white line from a digital signal have been proposed. For example, an image is predetermined by utilizing a brightness difference between the white line and a road surface. Method for obtaining white line candidate points by binarizing with a threshold of
As in Japanese Patent No. 3498, there is known a method of obtaining a white line candidate point by executing a rate of change of a brightness difference between a white line and a road surface, that is, a process of spatially differentiating a brightness signal.

Furthermore, there has been considered a method of obtaining a candidate point by time-integrating a signal obtained by subjecting an image signal to existing preprocessing such as binarization of the previous example or the image signal itself. The white line recognizing means for obtaining candidate points from such time integration will be described below.

FIG. 11 shows an example of an image when the front of the vehicle is photographed by the photographing means 1 which is a CCD camera.
Is a white line on the road, 5 is a road, and 6 is a character on the road. Here, the white line 4 is recognized for each scanning line, and the white line position is output for each of the left and right sides.

The configuration and operation for recognizing the white line from the input screen on one scanning line will be described below. FIG. 12 shows a flow (a) when a certain one scanning line (a line in FIG. 11) is extracted and processed, and the data (b) at that time is represented by the horizontal position on the screen. Is.

Reference numeral 8 in FIG. 12 (b) shows input image data when the vertical axis represents the luminance value. In this way, the white line 4, the characters 6 on the road, and the like have large luminance values, and the road 5 and the like have small luminance values. First, the contour line detection means 9 detects the input screen (input image data). The contour line detecting unit 9 obtains the contour line by the absolute value of the change amount of the input image data. The output data (contour detection data) of the contour detection unit 9 to which the input image data 8 is input is shown by a graph such as 10.

Next, the binarization means 11 performs binarization with a specific threshold value. When the contour detection data 10 is binarized with a specific threshold value, binarized data as shown by 12 is obtained. Next, the binarized data 12 is time-integrated by the time integration means 13.

For example, attention is paid to one pixel on the screen, and the time integration is constituted by a one-screen delay means 23 and an internal division means 24 as shown in FIG. The internal division means 24 receives two input τ− when given τ25 which defines an internal division ratio which is a certain time constant.
It outputs a 1: 1 internally divided value. The luminance value of the pixel at a certain time point is P (n), the time integrated value of the same pixel at the same time is I (n), and the time integrated value of the same pixel in the previous frame is I (n-1), which is a predetermined value. If the time constant of is τ, the time integration value is given by the following equation.

I (n) = (1-1 / τ) × I (n−1) + 1 / τ × P (n) (Equation 1)

Here, if the time constant τ is large, the time integration is long, and if it is small, the time integration is short. As a result, a contour line that is stable in time and appears at the same position strongly appears, and thus a contour line that is unstable in time such as a character on the road has a small value.

A graph obtained by time-integrating the binarized data 12 is as shown in FIG. In this way, the data corresponding to the position of the white line 4 has a large value, while the data corresponding to the position of the road character 6 has a small value.

Next, the data 1 obtained by the integrating means 13
4, the change amount detecting means 15 obtains change amount data 16. The change amount detecting means 15 obtains the change amount by the change amount absolute value similarly to the contour line detecting means 9.

Next, the scanning start point setting means 17 sets the center point on the line as the specific reference point 18 and sets it as the scanning start point. The scanning means 19 scans the variation data 16 in the left-right direction from the specific reference point 18. Further, in parallel with the scanning means 19, a change amount detecting means 20 detects a change amount in the change amount data 16 to detect a white line candidate point 21. The change amount detecting means 20 calculates, for example, a change amount of data and detects that the change amount data is equal to or more than a predetermined threshold value.

The white line candidate point 21 detected by the change amount detecting means 20 is determined by the judging means 22 as a white line position if the white line candidate point is stable in time series. Determine the position as the white line.

The white line position detecting means in one prior art of the present invention has been described above, but as a white line recognizing means in another prior art of the present invention, Japanese Patent Publication No.
As shown in Japanese Unexamined Patent Publication No. 24035, there is known a straight line group extraction unit that extracts a contour line from a captured image, performs Hough transformation, and extracts a straight line group that is approximate on the image.

[0017]

However, in such a conventional vehicle-mounted image processing apparatus, for example, a vehicle having a small relative speed to the host vehicle, a shadow provided on the road in parallel with the road, and the like are time-series. Since it is stable, it may be mistaken for a white line.

The present invention has been made to solve the above problems, and provides a vehicle-mounted image processing apparatus capable of performing stable white line recognition without being affected by noise which is stable in time series. The purpose is to

[0019]

According to a first aspect of the present invention, there is provided a vehicle-mounted image processing apparatus, a photographing means for photographing the periphery of a vehicle, an input means for inputting an image signal output by the photographing means, and the above-mentioned image. It is provided with a white line candidate point extracting means for extracting a white line candidate point based on a signal and a determining means for determining a white line from the extracted white line candidate points based on lane width information.

According to a second aspect of the present invention, there is provided an in-vehicle image processing device according to the first aspect, further comprising lane width learning means for learning the lane width based on the extracted white line candidate points. Be prepared.

An on-vehicle image processing apparatus according to a third aspect of the present invention is the vehicle-mounted image processing apparatus according to the second aspect, further comprising lane width learning condition setting means for learning the lane width under a predetermined condition. It is a thing.

A vehicle-mounted image processing device according to a fourth aspect of the present invention is the vehicle-mounted image processing device according to the second or third aspect, further comprising limiting means for limiting the amount of change in the lane width in learning the lane width. It is equipped with.

An on-vehicle image processing device according to a fifth aspect of the present invention is the on-vehicle image processing device according to any one of the first to fourth aspects, wherein time-integration processing is further performed on the extracted white line candidate point data. It is provided with a time integration means for performing.

An on-vehicle image processing device according to a sixth aspect of the present invention is the on-vehicle image processing device according to any one of the first to fifth aspects, further, the change amount of the white line position by the extracted white line candidate point. Is provided with a change amount limiting means for limiting.

An on-vehicle image processing device according to claim 7 of the present invention is the vehicle-mounted image processing device according to any one of claims 1 to 6, further using the obtained lane width,
A road type discriminating means for discriminating the road type is provided.

An on-vehicle image processing apparatus according to claim 8 of the present invention is the vehicle-mounted image processing apparatus according to any one of claims 1 to 7, further using the obtained lane width,
The vehicle body state determination means for determining the vehicle body state is provided.

An on-vehicle image processing apparatus according to claim 9 of the present invention is the vehicle-mounted image processing apparatus according to any one of claims 1 to 8, further using the obtained lane width,
It is provided with a road curvature detecting means for detecting a bend in the road.

[0028]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. FIG. 1 is a diagram showing an overall configuration of an embodiment of the present invention. In FIG. 1, 1 is an image capturing means (CCD camera), 2 is an A / D conversion means, 28 is a pre-processing means such as contour line detection, 29 is a time integration means, 30 is an area dividing means, 3 is a white line recognition means. , 45 is stabilization processing means,
Reference numeral 46 is a lane width learning means.

FIG. 2 shows an example of an image when the front of the vehicle is photographed by the image photographing means 1. This is an image obtained by performing preprocessing such as contour line detection and binarization processing on FIG. Is. FIG. 3 is a diagram when the image of FIG. 2 is time-integrated and divided into regions. In FIG. 3, 27 is a region I,
Reference numerals 26 denote areas O, respectively. Here, FIG. 2 described above is an output diagram of the preprocessing means 28, and FIG. 3 is the area dividing means 3
Output diagrams of 0 are shown respectively.

FIG. 4 shows an example of the preprocessing means. 31 is one scanning line delay means, 32 is one pixel delay means,
33 and 34 are difference means, 35 and 36 are absolute value means, 37
Is a specific threshold value, 38 and 39 are comparing means, and 40 is an OR operator. Here, the difference means 33 finds the change in the horizontal direction with respect to the camera, the absolute value means 35 finds the absolute value, and the comparison means 38 binarizes it to obtain the contour line in the horizontal direction with respect to the camera. On the other hand, the difference means 34 similarly obtains the vertical variation of the camera, the absolute value means 36 obtains the absolute value thereof, and the comparison means 39 performs binarization to obtain the contour line in the vertical direction with respect to the camera. Further, the OR operator 40 calculates the logical sum of the horizontal contour line and the vertical contour line to detect the contour line.

The contour-detected and binarized image is time-integrated by using the integrator shown in FIG. In this, a long time integration can be obtained by increasing the time constant τ in (Equation 1), and a short time integration can be obtained by decreasing the time constant τ. Therefore, for example, the part that corresponds to the short distance ahead of the road is relatively stable, so the time constant τ is set large, and the part that corresponds to the long distance ahead of the road is relatively stable, so the time constant τ is set small. The time constant may be properly used depending on the location.

The output signal from the time integrating means 29 is input to the area dividing means 30. This is performed by using the comparing means 42 as shown in FIG.
When the signal is larger than 1, the region I (27), the specific threshold 4
If it is smaller than 1, the screen is divided into two areas in the area O (26).

Here, as the specific threshold 41, for example, a value obtained by adding a specific value on the basis of the average of the entire screen or one scanning line of the signal integrated in time is used. For example, the time constant τ of the front part (short distance part) and the back part (long distance part) of the road
When the value of is changed, the threshold value is determined based on the average for each scanning line.

Area I of the areas divided by the preceding stage
Is an area where there is a contour line with little movement on the screen, and area O is an area where there is no movement of the contour line on the screen or the contour line itself.

For example, a white line, a vehicle, or the like that exists at the same position on the screen to some extent temporally becomes an area I near the area, and other areas, such as characters drawn on a road and a pedestrian crossing. Thus, when the vehicle is moving, the vicinity of what is not temporally present at the same place on the screen is
It becomes the area O.

In this way, by limiting the detection processing such as white line recognition or vehicle recognition to the region I, the processing speed can be increased as a whole and the stability of the apparatus can be improved. Next, the white line recognition means 3 performs white line recognition using this area. In this white line recognition, the white line in the area I is obtained from its contour.

FIG. 6 is a flowchart of white line recognition.
The white line recognition processing is performed for each scanning line, and further divided into two left and right from the center of the screen (specific reference point) and separately processed twice. First, for example, for the right side of the reference point, the flowchart of FIG. 6 is executed. First, it is checked whether or not the area I exists (step S1). If the area I does not exist, the previous white line recognition value is set as one of the candidate points (step S2). If the region I exists, scanning is performed in order from the region I closest to the center (step S3), and the presence of the contour line is determined (step S4). If it is determined that there is a contour line, up to two from the inside are set as candidate points for the white line recognition value (step S5), and the process proceeds to step S7. Step S7 will be described later. After the processing of step S2, or if there is no white line in the region I in step S5, the previous white line position (maximum value of the integrated value) is set as a candidate point for white line recognition (step S6), and step S7 is performed.
Proceed to. After the processing of step S7, the same processing is performed on the left side of the reference point.

Next, the process of step S7 will be described. In step S7, for example, with respect to the processing result on the right side, if there are contour line candidate points in step S5, regions (1) and (3) are designated (biased) as shown in the figure. On the other hand, steps S2 and S6
When the white line position is selected as the candidate point last time, the areas (2) and (4) are designated. Next, regarding the processing result on the left side, if there are contour line candidate points in step S5, areas (1) and (2) are designated, while the previous white line positions are set as candidate points in steps S2 and S6. If it does, areas (3) and (4) are designated.

Thus, in step S7, the left side,
When the respective designations on the right side are completed, the regions designated together on the left and right are selected as left and right white line candidate points, and white lines are determined as follows according to the regions (1) to (4).

The area (1) is a case where the area I exists on both the left and right sides and the contour line exists in the area I. In this case, of the candidate points, the one with the distance between the left and right candidate points closest to the lane width is determined as the white line recognition point.

In the area (2), the area I exists on the left side, and the contour line exists in the area I, and the area I does not exist on the right side, or even if it exists, the outline line exists in the area I. This is the case when it does not exist. In this case, the left candidate point is selected so that the distance from the previous right candidate point is closest to the lane width, and the right candidate point is determined so that the distance from the selected left recognition point is the lane width.

Right recognition point = left recognition point + lane width (when the right coordinate is larger than the left coordinate)

In the area (3), the area I exists on the right side, and the contour line exists in the area I, and the area I does not exist on the left side, or even if it exists, the outline line exists in the area I. This is the case when it does not exist. In this case, the right candidate point is selected so that the distance from the previous left candidate point is closest to the lane width, and the left candidate point is determined so that the distance from the selected right recognition point is the lane width.

Left recognition point = right recognition point-lane width (when the right coordinate is larger than the left coordinate)

The area (4) is the case where the area I does not exist on both the left and right, or the contour line does not exist in the area I even if it exists. In this case, of the left and right candidate points, the one having the closest left and right distance to the lane width is determined.
Here, the learned lane width is used as the lane width.
(Lane width learning will be described later).

Next, the stabilization processing means 45 performs stabilization processing on the recognition points. The stabilizing process is performed on the recognition point by using an integrating means and a variation limiting means. Here, the integrating means is obtained by the same method as the integrating means (FIG. 13) of the conventional device. The change amount limiting means sets the previous value of a certain value a (n) to a
(N-1) and the maximum change amount is δ (> 0),

When a (n) -a (n-1)> δ: a (n) = a (n-1) + δ

When -δ <a (n) -a (n-1) <δ: a (n) remains unchanged.

When -δ <a (n) -a (n-1) a (n) = a (n-1) -δ

The processing of is performed.

FIG. 7 shows a graph of the integration process and the change amount limiting process. In the figure, (a) shows an example when the integration processing is performed on the input data, (b) is an example when the variation limiting processing is performed on the input data. The time integration process has a function of stabilizing a small amount of noise, but has little effect on a large amount of noise.
On the other hand, the change amount limiting process has a function of stabilizing a large change amount of noise, but has a small effect on a small noise.

The lane width learning means 46 will be described below. Lane width learning is performed after the stabilization process, and learning is performed based on the left and right white line recognition point distances only when all of the following four conditions (a) to (d) are satisfied. Here, learning means recognizing and storing the lane width.

(A) In the case of the above area (1). (B) Left and right recognition point distance is ± 10% of standard lane width (on screen)
Within. (C) The distance between left and right recognition points is within ± 10% of the previous lane width. (D) The distance between the left and right recognition points is less than or equal to the lane width detected by the scanning line on the lower side of the screen (the actual positional relationship is in front).

As a result, only the lane width that meets the above conditions is reflected in the lane width, so that the correct lane width can always be obtained. The learning operation is performed by time integration of left and right recognition points. Thereby, the lane width can be stabilized in time. In addition, as the stabilizing operation, it is also possible to limit the amount of change as described above. In the above-described embodiment, steps A1 to S6 in the flowchart in the A / D conversion unit 2, the preprocessing unit 28, the time integration unit 29, the area dividing unit 30, and the white line recognition unit 3 are white line candidates of the present invention. Step S7 of the flow chart of the white line recognition means 3 constitutes the point extraction means, and constitutes the white line determination means of the present invention.

Embodiment 2 An example in which the type of traveling road can be determined from the lane width obtained by the above-described embodiment is shown below as another embodiment 2 of the present invention. For example, it is determined whether the road on the image is a general road or a highway based on the following conditions.

When lane width> 3.8 m Expressway When lane width ≤ 3.8 m General road

In this case, an integration process and a variation limiting process may be performed in order to stabilize the time series. In this way, the obtained information can be used to confirm the position of the own vehicle from the image by inputting it to the navigation or the like.

Embodiment 3 FIG. Next, an example of determining the state of the vehicle body from the lane width obtained by the above-described embodiment will be described below as a third embodiment. FIG. 8 is a view of the vehicle body traveling on the road as seen from above, and in FIG.
Is a white line, 54 is an angle formed by the vehicle body 56 and the white line 53, and 55 is a line when one scanning line on the screen is projected on the ground.

As shown in FIG. 8, when the angle between the vehicle body 56 and the white line 53 is θ, the recognized lane width is cose of the actual lane width.
cθ times. If the learned lane width is Wgakushu and the recognized lane width is Wninshiki, the vehicle normally runs parallel to the white line, and Wgakushu indicates the actual lane width value. Now, when the vehicle suddenly reaches an angle θ with the white line as shown in FIG. 8, Wninshiki is considered to be cosec θ times Wgakushu.

Cos ^-1 (Wgakushu / Wninshiki)

The angle between the vehicle body and the white line can be obtained by obtaining In addition, stability can be improved by increasing the number of lane width recognition points.

Embodiment 4 Next, an example in which the degree of bend of a road is determined from the lane width will be shown as a fourth embodiment. FIG. 9 is a view when the vehicle body is viewed from above. 56 is the vehicle body which is the own vehicle, 57 is the white line of the road, 58 is the center of the curvature circle of the road, 59 is the point A on the road, and 60 is on the road. Points B and 61 are an angle α formed between the A point 59 and the host vehicle 56, 62 is an angle β formed between the B point 60 and the host vehicle 56, 63 is a radius of curvature L, and 64 is between the A point 59 and the B point 60. Distance difference L in the direction of own vehicle
It is one.

Here, similarly to the case where θ is obtained in the above-described third embodiment, from the lane widths at the points A and B, the points A and B are obtained.
The angles α and β formed by the vehicle at the point and the white line can be obtained. Further, the distance difference L1 between the host vehicle and the host vehicle in the vehicle direction is known from the positional relationship of the cameras. Therefore, if the radius of curvature of the road is L,

Lsinβ-Lsinα = L1

Since the relationship of

L = L1 / (sin β-sin α)

Then, the radius of curvature is obtained.

An on-vehicle image processing apparatus according to claim 1 of the present invention is a photographing means for photographing the periphery of a vehicle, an input means for inputting an image signal output by the photographing means, and a white line candidate based on the image signal. Since the white line candidate point extracting means for extracting points and the determining means for determining the white line from the extracted white line candidate points based on the lane width information are provided, the time series is stable and reliable without being affected by noise. This has the effect of enabling highly recognizable white line recognition.

A vehicle-mounted image processing apparatus according to a second aspect of the present invention is the vehicle-mounted image processing apparatus according to the first aspect, further comprising lane width learning means for learning the lane width based on the extracted white line candidate points. Since it is provided, the effect of claim 1 can be further enhanced, and the white line can be recognized with high reliability.

The vehicle-mounted image processing apparatus according to claim 3 of the present invention is the vehicle-mounted image processing apparatus according to claim 2, further comprising lane width learning condition setting means for learning the lane width under a predetermined condition. The effect of claim 2 can be further enhanced, and the white line can be recognized with high reliability.

A vehicle-mounted image processing device according to a fourth aspect of the present invention is the vehicle-mounted image processing device according to the second or third aspect, further comprising limiting means for limiting the amount of change in the lane width for learning the lane width. Since it is provided, the effect of claim 2 or claim 3 can be further enhanced, and the white line recognition with high reliability can be performed.

An on-vehicle image processing device according to a fifth aspect of the present invention is the vehicle-mounted image processing device according to any one of the first to fourth aspects, further comprising time integration processing on the extracted data regarding the white line candidate points. Since the time integration means for performing the above is provided, there is an effect that these reliability can be further enhanced.

An on-vehicle image processing device according to claim 6 of the present invention is the on-vehicle image processing device according to any one of claims 1 to 5, further comprising a change amount of the white line position due to the extracted white line candidate point. Since the change amount limiting means for limiting the above is provided, there is an effect that the reliability of these can be further enhanced.

An on-vehicle image processing device according to claim 7 of the present invention is the vehicle-mounted image processing device according to any one of claims 1 to 6, further using the obtained lane width,
Since the road type discriminating means for discriminating the road type is provided, it is possible to perform the highly reliable road type discrimination.

An on-vehicle image processing device according to claim 8 of the present invention is the vehicle-mounted image processing device according to any one of claims 1 to 7, further using the obtained lane width,
Since the vehicle body state determination means for determining the vehicle body state is provided, it is possible to perform the vehicle body state determination with high reliability.

An on-vehicle image processing apparatus according to claim 9 of the present invention is the vehicle-mounted image processing apparatus according to any one of claims 1 to 8, further using the obtained lane width,
Since the road curvature detecting means for detecting the curve of the road is provided, the reliability of the detection of the curve of the road can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram of the present invention in a first embodiment of the present invention.

FIG. 2 is a diagram showing an example when an image of the front of the vehicle is subjected to preprocessing such as contour line detection in the first embodiment of the present invention.

FIG. 3 is a diagram showing an example when the image of FIG. 2 is subjected to time integration processing and divided into regions in the first embodiment of the present invention.

FIG. 4 is a diagram showing an example of preprocessing such as contour line detection in the first embodiment of the present invention.

FIG. 5 is a block diagram of area division according to the first embodiment of the present invention.

FIG. 6 is a flowchart showing a flow of white line recognition processing according to the first embodiment of the present invention.

FIG. 7 is a graph when time integration processing and change amount limiting processing are performed.

FIG. 8 is an explanatory diagram according to the third embodiment of the present invention.

FIG. 9 is an explanatory diagram according to the fourth embodiment of the present invention.

FIG. 10 is a block diagram of a conventional technique.

FIG. 11 is a diagram showing an example of an image obtained by capturing an image of the front of the vehicle in the related art.

FIG. 12 is a diagram showing a flow of recognition processing in a conventional technique.

FIG. 13 is a block diagram showing a configuration of time integration processing.

[Explanation of symbols]

1 image capturing means (CCD camera), 2 A / D conversion means, 3 white line recognition means, 4 white lines, 5 roads, 6 road characters, 7 background, 23 one screen delay means, 24 internal division means, 25 time constant τ, 26 area O, 27 area I, 28 preprocessing means, 29 time integration means, 30 area dividing means, 31 one scanning line delay means, 32 one pixel delay means, 33, 34 difference means, 35, 36 absolute value means , 37 specific threshold value, 38, 39 comparison means, 40 OR
Operator, 41 specific threshold, 42 comparison means, 53 white line,
54 an angle between the vehicle direction and the white line, 55 a line obtained by projecting one scanning line on the screen on the ground, 56 own vehicle, 57 white line,
58 Center of curvature circle of road, 59 Point A on road 60
Angle B between the point B and 61 A on the road and the vehicle α, 6
2 An angle β between the point B and the vehicle, 63, a radius of curvature L, 6
4 Distance in the host vehicle direction between points A and B.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 7/00 9061-5H G06F 15/70 330G

Claims (9)

[Claims] 1. A photographing means for photographing the periphery of a vehicle, an input means for inputting an image signal output by the photographing means, a white line candidate point extracting means for extracting a white line candidate point based on the image signal, and an extraction. An image processing apparatus for a vehicle, comprising: a deciding unit that decides a white line based on the lane width information from the selected white line candidate points. 2. The in-vehicle image processing device according to claim 1, further comprising lane width learning means for learning the lane width based on the extracted white line candidate points. 3. The vehicle-mounted image processing apparatus according to claim 2, further comprising lane width learning condition setting means for learning the lane width under a predetermined condition. 4. The in-vehicle image processing device according to claim 2, further comprising a limiting means for limiting a variation amount of the lane width in learning the lane width. 5. The vehicle-mounted image processing apparatus according to claim 1, further comprising a time integration unit that performs time integration processing on the extracted white line candidate point data. . 6. The in-vehicle image processing device according to claim 1, further comprising a change amount limiting unit that limits an amount of change in the white line position due to the extracted white line candidate points. . 7. The in-vehicle image processing apparatus according to claim 1, further comprising a road type determining unit that determines a road type by using the obtained lane width. apparatus. 8. The in-vehicle image processing apparatus according to claim 1, further comprising a vehicle body state determination means for determining a vehicle body state using the obtained lane width. apparatus. 9. The vehicle-mounted image processing apparatus according to claim 1, further comprising a road curvature detection unit that detects a bend in the road using the obtained lane width. Processing equipment.